Introduction

Sex differences are observed at several biologic levels, such as in brain anatomy and functioning, behavior, and susceptibility to neuropsychiatric or neurological disorders [1,2,3].

Due to the role played by biological sex in the anatomy, function and pathology of the human brain, the evaluation of its impact may provide valuable insights into the pathophysiological processes and progression of several neurological diseases.

Multiple sclerosis (MS) is the most common chronic, inflammatory, demyelinating and neurodegenerative disease of the CNS [4] that is characterized by heterogeneous incidence, prevalence, disease course, severity, and prognosis according to sex [5]. Females have a two to three times higher incidence and prevalence of the disease, a higher relapse rate throughout the disease and experience more frequently sensory symptoms [5]. Conversely, male MS patients present more frequently motor symptoms, and generally show a worse prognosis since their disease course is more frequently progressive, rapidly-evolving and characterized by more severe evolution [6] and disability accumulation [4,5,6]. No clear sex-based differences in response to disease modifying therapies (DMTs) have been documented, since most of DMT trials in MS have been conducted without sex-specific analyses of efficacy or adverse events. Some studies have occasionally shown sex-specific differences in the effects of DMTs. However, the limitation of the subgroup analysis design made it difficult to draw conclusions about the direction or magnitude of effect based on sex [7, 8].

In addition to clinical differences, pathophysiological processes underlying the sex-related differences in MS patients were also investigated using different MRI measures. Female MS patients showed a higher number of gadolinium-enhancing lesions [6, 9] and more significant white matter (WM) atrophy [10], suggesting a more inflammatory disease. Conversely, neurodegenerative processes seem to be more substantial in male MS patients, since they showed a significantly higher prevalence of T1-hypointense [9] and cortical lesions [11], more extensive and severe WM microstructural abnormalities [12] as well as global [12, 13] and gray matter (GM) atrophy [10, 12, 13], especially of deep GM nuclei [13].

Cognitive impairment is a frequent clinical manifestation in MS, with prevalence ranging between 40 to 70% [14, 15], involves several domains, occurs from the earliest phases of the disease, and becomes more prevalent and severe in progressive MS patients [14, 15]. Sex may influence also cognitive functions in MS [12, 13, 16,17,18,19,20,21]. Compared to females, male MS patients seem to be more impaired on several cognitive domains, including verbal memory [12, 13, 16, 17, 19], executive functions [12, 13, 18], attention [19], memory [12, 13, 17], visuospatial processing [17, 19], and information processing speed [12, 13, 19].

Differences of structural brain damage between female and male MS patients have been considered to explain sex-related differences in cognitive performances, although with conflicting findings [6, 9,10,11,12,13, 19, 22,23,24,25]. So far, more severe normal-appearing [NA] WM microstructural abnormalities [12] and subcortical GM atrophy [13] in male compared to female MS patients have been suggested as two relevant pathological substrates contributing to sex-related differences in cognitive impairment in MS. However, these results were not confirmed by other studies, possibly due to heterogeneities of the MS cohorts studied, neuropsychological tests administered, criteria applied to define cognitive impairment and MRI analyses performed [10, 22,23,24,25].

A multiparametric MRI approach is likely to improve our understanding of the underlying pathological substrates of sex-related cognitive differences in MS patients. To this aim, we explored whether differences in the regional distribution of focal WM lesions, NAWM microstructural abnormalities and GM atrophy may explain sex-related differences of cognitive performances in a large and well-characterized cohort of MS patients.

Materials and methods

Standard protocol approvals, registrations, and patient consents

Approval was received from the institutional ethical standards committee on human experimentation of IRCCS Ospedale San Raffaele for any experiments using human subjects (Protocol N° 2009-74). Written informed consent was obtained from all subjects prior to study participation according to the Declaration of Helsinki.

Subjects

From the database of Neuroimaging Research Unit, IRCCS San Raffaele Scientific Institute (Milan, Italy), we retrospectively selected 287 consecutive MS patients (173 females, 114 males). To be included, patients had to be ≥18 years old, have a diagnosis of MS according to the 2017 revised McDonald criteria, had to be relapse- and steroid-free for at least 3 months before study entry, have no significant neurological (other than MS) or psychiatric condition that could interfere with cognitive functioning, and have a stable treatment for MS from at least 6 months. In addition, 172 healthy controls (HC) (92 females, 80 males), who underwent the same MRI protocol, and had no neurological diseases or systemic disorders potentially affecting the central nervous system and a completely normal examination were selected.

Clinical and neuropsychological assessment

On the day of MRI acquisition, all MS patients underwent a complete neurological examination, with rating of the Expanded Disability Status Scale (EDSS) and a neuropsychological evaluation using the Brief Repeatable Battery of Neuropsychological Tests (BRB-N) [26]. In particular, we used the Selective Reminding Test to assess verbal memory, the 10/36 Spatial Recall Test to assess visual memory, the Symbol Digit Modalities Test and the Paced Auditory Serial Addition Test 2” and 3” to assess information processing speed (IPS)/attention, and the Word List Generation to assess verbal fluency. Z-scores for all BRB-N tests were calculated using sex-, age- and education-adjusted regression models according to the normative data from an Italian representative sample [27]. Then z-scores for specific cognitive domains were quantified by averaging z-scores of tests belonging to the previous cognitive domains, as previously described [28]. Finally, z-scores of global cognitive functions (obtained by averaging z-scores of cognitive domains) were quantified [28]. Test failure was defined as a score below the fifth percentile of normative values [29]. Impairment in a cognitive domain was defined as failure in at least one test of BRB-N assessing that domain and cognitive impairment was defined as impairment in at least two cognitive domains [29].

MRI acquisition

Using the same 3.0 Tesla Philips Intera scanner (Philips Medical Systems, The Netherlands), the following sequences of the brain were acquired from all subjects during a single session: (a) dual-echo (DE) turbo spin echo (TSE) (TR/TE = 2599/16–80 ms; flip angle = 90°; FOV = 240 mm2; matrix = 256 × 256; echo train length = 6; slice thickness = 3 mm, 44 contiguous axial slices); (b) 3D T1-weighted fast field echo (FFE) (TR/TE = 25/4.6 ms, flip angle = 30°, FOV = 230 mm2, matrix = 256 × 256, slice thickness = 0.9 mm, 220 contiguous axial slices, in-plane resolution = 0.89 × 0.89 mm2); (c) pulsed-gradient spin echo EPI (TR/TE = 8938/58 ms, acquisition matrix = 112 × 88, FOV = 240 × 231 mm2, slice thickness = 2.3 mm, 55 contiguous axial slices) with SENSE (acceleration factor = 2) and diffusion gradients applied in 35 non-collinear directions. Two optimized b factors were used: b1 = 0, b2 = 900 s/mm2. The slices of all the scans were positioned to run parallel to a line that joints the most infero-anterior and infero-posterior parts of the corpus callosum.

Conventional MRI analysis

T2-hyperintense white matter (WM) lesion volumes (LV) were quantified on DE sequence using a local thresholding segmentation technique (Jim 7.0, Xinapse Systems Ltd, Colchester, UK) [30, 31]. After T1-hypointense lesion refilling, normalized brain volume (NBV), GM volume (NGMV) and WM volume (NWMV) were calculated on 3D T1-weighted images using the SIENAx software.

T2-hyperintense WM lesion probability maps

For each MS patient, binarized masks of T2-hyperintense WM lesions were obtained, rigidly co-registered to the 3D T1-weighted scans and normalized to the standard space (using the Diffeomorphic Anatomical Registration using Exponentiated Lie algebra [DARTEL] non-linear transformation). Finally, they were smoothed with an 8-mm Gaussian kernel and averaged to obtain T2 lesion probability maps (LPM).

Diffusion-weighted MRI analysis

Diffusion-weighted images were first corrected for distortions caused by eddy currents and for subject’s movement (http://web.stanford.edu/group/vista/cgi-bin/wiki/index.php/DTI_Preprocessing#dtiRaw_Preprocessing_Pipeline) [32]. In details, we applied an algorithm that combines 3D rigid transformation for motion correction with a constrained non-linear warping for eddy-current distortions. The Rohde model parameters are estimated by maximizing the normalized mutual information between each diffusion-weighted image and the non-diffusion weighted image mean. Then, using the FMRIB’s Diffusion Toolbox (FDT tool, FSL 5.0.5), the diffusion tensor was estimated in each voxel by linear regression and fractional anisotropy (FA) maps were obtained.

Tract-based spatial statistics (TBSS) analysis was used to perform a voxel-wise analysis of the whole-brain WM FA (http://www.fmrib.ox.ac.uk/fsl/tbss/index.html/). Individual FA images were non-linearly registered to the FMRIB58_FA atlas provided within FSL and averaged to obtain a customized atlas. Then, the resulting FA atlas was thinned to create a WM tract ‘skeleton’, which was thresholded at a FA > 0.2 to include WM voxels only. Individual-subject’ FA values were projected onto this group skeleton by searching perpendicular from the skeleton for maximum FA values. Maximum FA values were chosen to restrict analysis to the center of WM tracts (where maximum FA values are found), rather than considering voxels at the edge of tracts, which may suffer from partial volume effects.

Voxel-based morphometry analysis

Voxel-based morphometry (VBM) and DARTEL, as implemented in SPM12 (www.fil.ion.ucl.ac.uk/spm/), were used to map regional GM volume differences among study groups. The lesion-filled 3D T1-weighted images were used for a group-wise alignment. First, the images were segmented into the different tissues using the standard segmentation procedure of SPM12. Then, the segmented GM and WM images of all subjects were used to create GM and WM templates. At each iteration, the deformations calculated using the DARTEL registration method were applied to GM and WM with an increasingly good alignment of subject morphology, to produce templates. Finally, an affine transformation was computed that maps from the population GM template to the MNI space. GM maps were spatially warped, modulated and smoothed with an 8 mm Gaussian kernel.

Statistical analysis

Demographic and clinical data were compared between groups, according to sex (female and male) and the disease status (HC and MS), using chi-square, two-sample t, or Mann-Whitney test. Sex-related differences in MS patients’ cognitive performances were assessed by comparing z-scores, at test, domain and global level, with two-sample t-tests. Prevalences of cognitive impairment were also reported. Age-adjusted linear models for heteroscedastic data were applied to analyze conventional MRI variables (i.e., brain T2-hyperintense WM LV and global volumetric measures). Brain T2-hyperintense WM LV were log-transformed. False discovery rate (Benjamini–Hochberg procedure) correction was carried out to take into account the overall number of tests performed. A p < 0.05 was considered statistically significant.

Voxel-wise differences in T2-LPMs, NAWM FA values (TBSS), and regional GM volumes (VBM) according to sex, disease status, and their interaction were explored using age-adjusted linear models. Voxel-wise associations of MRI variables with cognitive performances (global and cognitive domain z-scores) in female and male MS patients were investigated by age-adjusted linear models. Sex-related differences of association were also tested. T2-hyperintense WM lesion masks was also included as voxel-wise covariate in TBSS analyses to focus only on NAWM, whereas the inverse SIENAX v-scaling factor was included as additional covariate in GM volume analyses to correct for head-size.

For WM lesions and GM volumes, the above-described analyses were conducted using SPM12 (http://www.fil.ion.ucl.ac.uk/spm/software/spm12/). Inclusion masks obtained from the WM (for lesions analyses) or GM (for GM volumes analyses) DARTEL templates, transformed to the MNI space, smoothed and thresholded at 0.25, were applied. For TBSS data, a permutation method (FSL 6.0.1 “Randomise” program) was used with 5000 permutations.

Results of voxel-wise analyses were assessed at p < 0.01 (family-wise error [FWE]-corrected for multiple comparisons). Moreover, for T2-LPMs and VBM analyses a cluster extension (kE) threshold of 10 voxels was applied.

Results

Demographic, clinical and conventional MRI measures

Male HC had significantly lower NBV (p < 0.001) and NGMV (p < 0.001) compared to female HC. Female and male MS patients had significantly lower years of education, NBV, NGMV and NWMV and higher T2-hyperintense WM LV compared to female and male HC (all p ≤ 0.029) (Table 1). Female MS patients had significant lower years of education compared to male MS patients (p = 0.043).

Table 1 Demographic, clinical and conventional MRI features of the study groups.
Full size table

Seventy MS patients received no DMT (female MS patients = 45/173 [26%]; male MS patients = 25/114 [22%]), 130 MS patients received first-line DMTs (female MS patients = 73/173 [42%]; male MS patients = 57/114 [50%]) and 87 MS patients received second-line DMTs (female MS patients = 55/173 [32%]; male MS patients = 32/114 [28%]), with no between-group differences (p = 0.422) (Table 1).

Neuropsychological findings

Table 2 summarizes the results of cognitive assessment in MS patients according to sex. Sixty-six female (38%) and 48 male (42%) MS patients were cognitively impaired, with no significant between-group difference in the prevalence of cognitive impairment (p = 0.704). The prevalence of impairment in each cognitive domain was similar between female and male MS patients. However, for verbal memory, a higher prevalence of impairment (p = 0.008) and worse performance (p = 0.001) were found in male compared to female MS patients. Conversely, females performed significantly worse than males in verbal fluency (p = 0.004). No significant differences were found between female and male MS patients in z-scores of global cognitive functions, visual memory, and IPS/attention (all p ≥ 0.289).

Table 2 Mean z-scores and prevalence of impairment at BRB-N in MS patients, according to sex.
Full size table

T2-hyperintense WM LPMs

T2-hyperintense WM lesions were mostly located in the bilateral corona radiata and periventricular WM, with no significant differences between female and male MS patients (Fig. 1A).

Fig. 1: T2-hyperintense WM lesion probability maps.
figure 1

A T2-hyperintense WM lesion probability maps in (1) female MS patients and (2) male MS patients (red-yellow coded for frequency). B Voxel-wise negative associations between the prevalence of T2-hyperintense WM lesions and worse cognitive performances (global z-scores and of each cognitive domain) in (1) female and (2) male MS patients. Regions with a significantly stronger association in female compared to male (3) MS patients are also shown. The lesion probability maps are thresholded to show voxels in which lesion frequency of 5% is present, up to a maximum lesion frequency of 40%. Results (SPM12; p < 0.01 FWE-corrected, kE = 10; red-yellow coded for t value) are superimposed on the customized GM template. Images are presented in neurological convention. A anterior, FWE family-wise error; IPS information processing speed, kE cluster extension, L left, MS multiple sclerosis, P posterior, R right, SPM statistical parametric mapping, WM white matter.

Full size image

Significant correlations were found between the frequency of T2-hyperintense WM lesions in clinically-relevant WM tracts and worse cognitive performances. Specifically, in female MS patients, a higher frequency of T2-hyperintense WM lesions in the corpus callosum, corona radiata, bilateral anterior thalamic radiation (ATR), forceps major, bilateral superior longitudinal fasciculi (SLF), bilateral inferior longitudinal fasciculi (ILF) and bilateral inferior fronto-occipital fasciculi (IFOF) was significantly associated with worse performances in global cognitive function as well as in each cognitive domain explored, except for verbal fluency (all p ≤ 0.006) (Fig. 1B). Conversely, in male MS patients a higher frequency of T2-hyperintense WM lesions in corpus callosum, bilateral ILF and IFOF, left ATR and forceps major was significantly associated with worse performance in global cognitive functions, verbal memory and IPS/attention (all p ≤ 0.006), with no associations found in visual memory and verbal fluency. A stronger association between T2-hyperintense WM lesions in left ATR (p = 0.002) and SLF (p = 0.004) with worse performance in global cognitive functions was found in female compared to male MS patients (Fig. 1B).

WM microstructural abnormalities

In HC, no significant differences in NAWM FA values between females and males were found. Compared to sex-matched HC, male and female MS patients showed significantly lower NAWM FA values in the majority of brain WM tracts (for both female and male MS patients all p < 0.01) (Fig. 2A). A significant interaction between sex and disease was found, with a larger effect of MS on lowering NAWM FA values in left cingulum, SLF, anterior and superior corona radiata and body of corpus callosum occurring in females compared to males (all p < 0.01) (Fig. 2A).

Fig. 2: WM microstructural abnormalities.
figure 2

A Voxel-wise FA values differences between different study groups: (1) significantly lower FA in female MS patients vs female HC; (2) significantly lower FA in male MS patients vs male HC; (3) Significantly larger effect of MS on lowering FA in female compared to male MS patients (significant sex × disease interaction effect). B Voxel-wise positive associations between lower FA and worse cognitive performances (global z-scores and of each cognitive domain) in (1) female and (2) male MS patients. Regions with a significantly stronger association in female compared to male (3) MS patients are also shown. Results (TBSS) were thresholded at p < 0.01 FWE-corrected; red-yellow coded for p-value. The WM skeleton, thresholded at FA > 0.2 and represented in green, is superimposed on the mean FA map derived from all the subjects. Images are presented in neurological convention. A anterior, FA fractional anisotropy, FWE family-wise error, HC healthy controls, IPS information processing speed; L left, MS multiple sclerosis, P posterior, R right, TBSS tract-based spatial statistics, WM white matter.

Full size image

Significant positive associations were found between cognitive performances and NAWM FA values in both female and male MS patients, with some differences. In female MS patients, widespread lower NAWM FA values were significantly associated with worse performances in global cognitive functions as well as in each cognitive domain explored (all p < 0.01). In male MS patients, significant associations were found between lower NAWM FA values in the majority of WM tracts and worse performances only in global cognitive functions and in IPS/attention (all p < 0.01) (Fig. 2B). A stronger association between lower NAWM FA values in forceps minor and genu of corpus callosum and worse global cognitive performance (all p < 0.01) as well as between lower NAWM FA values in forceps major, splenium of corpus callosum, right SLF and worse verbal memory performance (all p ≤ 0.007) were found in female compared to male MS patients (Fig. 2B).

GM atrophy

In HC, males showed a significantly lower GM volume in clusters localized in right postcentral gyrus, left rectus gyrus, left rolandic operculum and left anterior cingulum gyrus compared to females (all p ≤ 0.005) (Fig. 3A). Female and male MS patients showed a significant widespread GM atrophy compared to sex-matched HC (for female MS patients all p ≤ 0.008; for male MS patients all p ≤ 0.001) (Fig. 3A). No significant interaction between sex and disease was found.

Fig. 3: GM atrophy.
figure 3

A Voxel-wise GM volume differences between different study groups: (1) regions with lower GM volume in male compared to female HC; (2) regions with lower GM volume in female MS patients compared to female HC; (3) regions with lower GM volume in male MS patients compared to male HC. B Voxel-wise positive associations between regional GM volume loss and worse cognitive performances (global z-scores and of each cognitive domain) in (1) female and (2) male MS patients. Results (SPM12; p < 0.01 FWE-corrected, kE = 10; red-yellow coded for t values) are superimposed on the customized GM template. Images are presented in neurological convention. A anterior, FWE family-wise error; GM gray matter, HC healthy controls, IPS information processing speed, kE cluster extension, L left, MS multiple sclerosis, P posterior; R right, SPM statistical parametric mapping.

Full size image

In MS patients, several positive associations between regional GM volumes and cognitive performances were found. Specifically, in female MS patients worse performance in global cognitive functions and IPS/attention was associated with lower GM volumes in fronto-parietal lobes, insula, hippocampus, cingulate cortex, and deep GM (all p ≤ 0.003). Worse verbal and visual memory performance was associated with lower hippocampal and deep GM volumes (all p ≤ 0.006) (Fig. 3B). In male MS patients, a lower GM volume in bilateral temporal lobes, middle cingulate cortex and hippocampus, left precuneus, left posterior cingulate cortex was significantly associated with worse performances in global cognitive functions as well as in each cognitive domain explored (all p ≤ 0.007) (Fig. 3B). No association between performance in verbal fluency and regional GM volume was found for either female or male MS patients. No significant sex-related differences of associations between regional GM atrophy and cognitive performances were found.

Discussion

By evaluating a large and well-characterized cohort of MS patients, no significant sex-related differences in global cognitive functions nor in the prevalence of being cognitively impaired were found, whereas verbal memory performance was worse in male MS patients and verbal fluency performance was worse in female MS patients. Focal WM T2-hyperintense lesions, NAWM microstructural abnormalities and GM atrophy contributed to explain cognitive dysfunction in both male and female MS patients, with some sex-related differences. Worse cognitive performance was more strongly associated with focal WM lesions and NAWM microstructural abnormalities in female MS patients. Deep GM volume loss seemed to be associated with worse cognitive performances especially in female MS patients, whereas, in male MS patients, cortical GM volume loss was associated with worse cognitive performances, although without significant between-group differences.

In our cohort, a substantial proportion of MS patients (38% of females and 42% of males) were cognitively impaired, but no significant between-group differences were found regarding the prevalence of impairment nor the global cognitive performance. This proportion is slightly lower than that reported in literature [14]. However, there is no unambiguous criterion to define cognitive impairment in MS patients, and this causes great variance between different studies [15]. Since neural correlates are likely to differ across specific cognitive domains [15], we decided to use criteria that are more conservative but at the same time more focused on separate cognitive domains to define cognitive impairment [29].

The possible influence of sex on cognitive impairment in MS is still a matter of investigation [12, 13, 16,17,18,19,20,21]. Previous studies suggested that male MS patients may be characterized by more prevalent and severe cognitive impairment [13, 16,17,18,19], but with heterogeneous and conflicting findings. Worse cognitive performance in male MS patients has been frequently found compared to HC, but more rarely directly to female MS patients [12]. Moreover, cognitive performance has been frequently compared in a limited subset of neuropsychological tests or cognitive domains, whereas, global cognitive performance was explored only in a few studies [12, 13], where female MS group showed little [12] or no significant cognitive dysfunction [13]. These conflicting findings may be explained by heterogeneities of the MS cohorts studied (i.e., severity of clinical disability, proportion of different clinical phenotypes, and disease duration), neuropsychological tests administered, and criteria applied to define cognitive impairment [10, 22,23,24,25].

MS cohort showed that IPS/attention and verbal memory were the most frequent impaired domains, confirming that these cognitive deficits are the most common dysfunction in MS patients of both sexes [14, 15]. Male MS patients showed worse performance and a significantly higher prevalence of impairment in verbal memory. Our results are consistent with previous studies that showed a significant frailty in verbal memory abilities in male compared to female MS patients [12, 13, 16]. On the other hand, unexpectedly, female MS patients showed worse performance in verbal fluency, although without differences with male MS patients in the prevalence of being impaired. Verbal abilities has been found to favor females [16, 33, 34]. Even though the association between brain structural damage and verbal fluency was not significantly different according to sex, it is tempting to speculate that in females this domain may be particularly vulnerable or more sensitive to the accumulation of MS-related structural damage. Clearly, further studies are necessary to confirm our results.

In MS patients of both sexes, worse cognitive performance was associated with a higher prevalence of focal T2-hyperintense WM lesions in several cognitively-relevant WM tracts, such as ATR, forceps minor/major, SLF, ILF and IFOF. The severity of damage within focal T2-hyperintense lesions may also contribute to the disruption of critical WM pathways by anterograde/trans-synaptic degeneration secondary to axonal transection [35]. Thus, damage of these WM tracts, connect cortico-subcortical GM structures and are involved in several cognitive functions, has been consistently associated with worse cognitive performances in MS [12, 35,36,37,38,39,40,41,42,43]. Our findings support the role of a disconnection syndrome as one relevant pathological mechanisms of cognitive impairment in these patients.

When we explored sex-related differences in the strength of associations, focal T2-hyperintense WM lesions in left ATR and SLF were found to be more associated with worse global cognitive functions in female compared to male MS patients. These results suggest a more relevant contribution of focal inflammatory demyelinating lesions to worse cognitive performance in female MS patients, possibly due to sex-related differences in the immune-pathophysiology of the disease and due to the different effects of sex hormones in males and females [5, 6, 9, 44].

The analysis of NAWM microstructural abnormalities showed lower FA values in the majority of WM tracts in both female and male MS patients compared to their sex-matched HC. The interaction between sex and disease showed that the effect of MS in determining NAWM microstructural abnormalities may be more relevant in females compared to males.

In both female and males MS patients, lower FA values in the majority of WM tracts were significantly associated with worse performance in global cognitive functions and in each cognitive domain explored for females MS patients, whereas only in global cognitive functions and IPS/attention for male MS patients. These results confirm that diffuse NAWM damage may also undermine the physiological connections among cognitively relevant GM regions [12, 14, 36, 37, 39, 45, 46], explaining cognitive deficits in MS patients of both sexes. Interestingly, a stronger association between lower FA values in several WM tracts and global cognitive function and verbal memory performance was found again in female compared to male MS patients. This result further supports the hypothesis that pathological processes that affect the NAWM, including inflammation, with an increased density of activated microglia and lymphocytes, demyelination, astrogliosis and axonal damage, may contribute to cognitive impairment in female more than in male MS patients [5, 6, 9, 44].

Our results seem conflicting with previous findings showing no sex-related differences [10] or more severe microstructural abnormalities in male compared to female MS patients [12]. However, one study did not evaluate brain diffusivity abnormalities at a voxel-level [10], whereas another [12] explored sex-related microstructural abnormalities in a cohort mainly including relapsing-remitting MS patients, with mild disability and short disease duration. Accordingly, it could be possible that, in the early phases of the disease, WM diffusivity abnormalities may be more severe in male compared to female MS patients [12]. Conversely, in MS patients with different clinical phenotypes, longer disease duration and more severe disability, the inflammatory pathophysiological processes characterizing female more than male MS patients become more evident.

Regarding regional GM volume, the VBM analysis revealed that male HC showed significantly lower NGMV and lower GM volumes in frontal regions compared to female HC. Nevertheless, such findings are debated since other studies found higher GM volumes in males [10]. The evaluation of heterogeneous HC cohorts, the adjustment for different covariates (intracranial volume, height and weight), and the evaluation of different GM measures (i.e., global or regional GM volumes, percentage of GM volume, cortical thickness) may explain discrepancies among studies [3, 47, 48].

As expected, MS patients of both sexes showed a widespread pattern of GM atrophy compared to sex-matched HC. However, the analysis of interaction between sex and disease showed that no voxel was statistically significant. In both male and female MS patients, worse cognitive performances, except for verbal fluency, was significantly associated with lower GM volume in several cortico-subcortical GM regions, confirming the relevance of atrophy in cognitively-relevant GM regions as another relevant contributor of cognitive impairment in both female and male MS patients [13, 37, 38, 49, 50]. Specifically, our results supported the relevance of deep GM, hippocampus, and other cognitively relevant cortical regions of fronto-temporo-parietal lobes in explaining cognitive dysfunction in MS patients [37, 49,50,51,52,53,54]. Notably, although no significant sex-related differences of associations between regional GM atrophy and cognitive performance were found, worse cognitive performance seemed to be associated mainly with deep GM volume loss in female MS patients, and with cortical GM volume loss in male MS patients. Although further studies are needed to confirm our findings, it is tempting to speculate that, at least in our cohort of MS patients, a more inflammatory pathophysiology affecting WM tracts of female MS patients may promote a cognitively-relevant deep GM atrophy. Conversely, primary damage of cortices involved in cognitive functions may be a relevant contributor of worse cognitive performances in male MS patients.

Our results seem in contrast with previous studies investigating sex-related differences in regional GM volume that found more severe GM atrophy in males, compared to female MS patients [12, 13] especially affecting the deep GM [12]. However, again, the studies mainly focused on a cohort of mainly early relapsing-remittent MS patients, with a short disease duration and mild disability. The more severe abnormalities found in the early stages of the disease on diffusion metrics in male MS patients [12] may also explain the sex-related differences in regional GM volume found, particularly with regard to deep GM volume. This would be in line with several studies showing an early occurrence of deep GM atrophy as a consequence of the occurrence of focal T2-hyperintense WM lesions and microstructural NAWM damage, due to the central and highly connected nature of these regions [55,56,57]. Conversely, the cohort of MS patients evaluated in our study is characterized by a more long-standing disease course. Accordingly, a possible sex effect that is present in the early phases of disease may be less evident, whereas a more prominent role of cortical atrophy in determining cognitive impairment may emerge [50].

Our study is not without limitations. Cortical lesions have been associated with cognitive deficits in MS patients, but specific MRI sequences for their quantification were not available in our cohort of MS patients. The cross-sectional setting of this study did not allow to explore the contribution of the pathological substrates to cognitive impairment according to sex in the different phases of the disease. Levels of sex hormones were not available, thus we could not explore their influence in the associations between cognitive functions and brain structural damage. The four groups evaluated in the present study differed on some clinical (i.e., educational level) and volumetric MRI variables, which may also contribute to the results obtained. Further studies are needed to confirm that our results are not specific for the population of the present study. Moreover, our HC did not perform neuropsychological assessment, thus further studies are needed to confirm that the results of the present study are MS-specific. However, it should be noted that we corrected and standardized the MS patients’ cognitive scores according to Italian normative data, calculated on a relevant cohort (n = 200) of healthy controls [27]. Finally, BRB-N did not have test assessing social cognition, thus further studies are needed to evaluate this understudied cognitive function in MS and assess if sex-related differences may occur.

In conclusion, in this well-characterized cohort of male and female MS patients the prevalence and severity of cognitive impairment are similar, with only limited sex-related differences in the involvement of specific cognitive domains. In MS patients of both sexes, focal WM lesions, NAWM microstructural abnormalities, and GM volume loss are relevant contributors of cognitive impairment. However, sex influences patterns of NAWM microstructural abnormalities and associations between focal T2-hyperintense WM lesions, NAWM microstructural abnormalities, and cognitive performance. Specifically, a disconnection syndrome caused by the accumulation of focal T2-hyperintense WM lesions and NAWM microstructural abnormalities may be more relevant in female MS patients to explain cognitive impairment.